SCD4X CO2 Sensor Driver & Docs

[mbz4]

Extend support to include the Sensirion SCD40 compact photoacoustic NDIR CO2 sensor.

Tasks:

• understand features, examples,
• test available drivers,
• write, validate IoTempower bridge,
• document new driver,
• merge PR.

Bare sensor module:

Module in use for this driver development and in local stock:

Resources:

• git: embedded i2c scd4x driver
• RPi i2c scd4x driver
• platformio registry: scd4x drivers
• datasheet: scd4x
• aliexpress: scd40 i2c module board ~20-30eur (expected 70eur)
• Promising - PlatformIO: DFRobot_SCD4X
• Adafruit SCD-40 - True CO2, Temperature and Humidity Sensor - STEMMA QT / Qwiic

[mbz4]

Update: ended up testing daisy chained with SGP30, SHT30, ENS160 (not sure my unit works) and AHT2X sensors together on the same board and same platformio project, felt convenient as the SCD4x official i2c library has a few calibration and offset features so it made sense to merge the project; this way the script has simultaneous access to all the measurement feeds for easier printouts and testing different libraries

[mbz4]

• Started with this library: git: DFRobot_SCD4X
• most promising alternative: platformio: scd4x

Need to extend IoTempower basic wrapper support first, making it possible to configure a simple (calibrated) measurement cycle. Afterwards document the basic features and finally focus on exposing advanced features. In the delivered stage, the sensor should be fully configurable through setup.cpp and reconfigurable through MQTT/sensor_topic....

[mbz4]

Tried available libraries for async measurement support but the sensor is not robust (plug/unplug test - measurements differ too much, unreliable). Digging deeper into sensirion scd4x library...

[mbz4]

There be whales here! 🥇

[ulno]

Maybe link the bias ribbit finding here? Maybe it could be solved with an IoTempower filter? What about an initial PR to encourage more testers?

[mbz4]

Maybe link the bias ribbit finding here?

GMAC2024-_Ribbit_NetworkPoster-_White_BG_48_x_36_in.pdf

• Ribbit issue detailing their decision process narrowing down sensor options: https://github.com/Ribbit-Network/ribbit-network-frog-hardware/issues/103

Maybe it could be solved with an IoTempower filter?

ribbit in the poster don't really state it as a linear steady state bias error... so do we just need to the add the difference to each measurement to offset the error from reference sensor (vaisala)? ... i think we need to test this over a variety of temp, humidity ranges + CO2 concentration levels to really show the behaviour; but the datasheet does show deviations in specific ranges for the scd4x: https://sensirion.com/media/documents/48C4B7FB/64C134E7/Sensirion_SCD4x_Datasheet.pdf

What about an initial PR to encourage more testers?

Can do, but will have to then regress the updated landing page architecture diagram. Will prepare draft for ENS160 first then start the PR (and new issue for that sensor).

[mbz4]

PR: https://github.com/iotempire/iotempower/pull/118